There will be approximately three steps to this tutorial.
- Install Homebrew
brewto install the latest version of
Verify that the installation is working correctly and troubleshoot if necessary.
- Install useful packages using
Ensure they are working and troubleshoot if necessary.
A main tool in our installation process will be homebrew. Homebrew is a package manager for Mac OS that is extremely handy, well beyond the scope of installing Python. First we’ll check if homebrew is already installed. To do so, run the following snippet and look examine the output. If homebrew is installed, then it should print out some version information.
> brew --version Homebrew 2.2.12 Homebrew/homebrew-core (git revision d51e0f; last commit 2020-04-09) Homebrew/homebrew-cask (git revision 78b7f; last commit 2020-04-09)
If running the above command gives an error because
brew is not a defined
command, then you’ll have to install homebrew. To do so, open a new Terminal
window and enter the following (this snippet was taken from the homebrew main
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install.sh)"
As a brief caveat, it’s possible you will first have to install some XCode
components required by
brew, in which case
brew should let you know. Be
prepared to run the following snippet and follow along with the install
If the installation of brew worked, then you should now be able to run
and see some version information as per the above.
Now, installing Python should be as easy as running:
> brew install python3 <<long>> > python --version Python 3.6.10 :: Anaconda, Inc. > pip -V pip 20.0.2 from /Users/aberk/anaconda/envs/py37/lib/python3.6/site-packages/pip (python 3.6)
Note: An alternative to the homebrew installation avenue is to download an installer file from the official Python page. I have no familiarity using this avenue.
Installing commonly used packages
This tutorial was tailored for a machine learning audience, so the installed packages will contain some packages that are useful for ML.
> pip install numpy pandas matplotlib > pip install scipy seaborn sckit-learn > pip install feather-format > pip install torch torchvision > pip install notebook jupyterlab
There are numerous other useful packages to install, but let’s see if we can get up and running with these first. In all cases, you should be able to confirm that a package was installed correctly by running the following snippet and observing the package’s version number as the output.
> python -c 'import the_package; print(the_package.__version__)'
For example, on my machine:
> python -c 'import numpy; print(numpy.__version__)' 1.18.1
In rare instances, a package may not implement the (more-or-less standard)
.__version__ syntax. Python will throw an error in this exceptional case.
See here for some alternative information on installing Python and managing environments. The post seems slightly out-of-date (by about a year). The short timeframe of this obsolescence does not bode well for the present tutorial. 😅
For learning how to program in Python, the official documentation provides extensive resources for getting up-and-running, as well as a language reference which serves as a bit of a dictionary once you’re mostly up-and-running. Another great tutorial resource is Patrick Walls’s UBCS3 page which includes some “mathematical Python” walkthroughs, and “machine learning with Python” walkthroughs.
Intro to PyTorch
The purpose for this post was to establish the environment for an Intro to PyTorch tutorial. Follow the link for the repo.